argoproj / argo-cd

Declarative Continuous Deployment for Kubernetes
https://argo-cd.readthedocs.io
Apache License 2.0
16.67k stars 5.05k forks source link

LoadBalancer not presenting on EKS #18812

Open ucfchandra opened 3 days ago

ucfchandra commented 3 days ago

Checklist:

Describe the bug

When I launch the application at github.com/ucf/ucf-announcements-django in the dev branch from the helm folder, it spins up the resources properly. However, the LoadBalancer is not accessible, and on AWS it shows 0/3 instances being spun up. Argo says that the resource is healthy though.

To Reproduce

Create an app from the github.com/ucf/ucf-announcements-django repo from the dev branch using the /helm folder.

Expected behavior

Should be able to go to the URL provided by the service and see the application running.

Screenshots

Version

# Paste the output from `argocd version` here.
argocd: v2.11.3+3f344d5
  BuildDate: 2024-06-06T12:42:22Z
  GitCommit: 3f344d54a4e0bbbb4313e1c19cfe1e544b162598
  GitTreeState: clean
  GoVersion: go1.22.4
  Compiler: gc
  Platform: darwin/amd64

Logs

Paste any relevant application logs here.

Can't find any error logs to debug

agaudreault commented 2 days ago

Hey @ucfchandra, This seems to be a configuration issue with the chart. Argo CD will only synchronize the resources in the destination namespace, a little bit like you would do manually with kubectl. If the LoadBalancer is not created properly, it may be caused by misconfiguration of external components like https://kubernetes-sigs.github.io/aws-load-balancer-controller/ responsible to create the LoadBalancer.

Argo Health is based on individual resources health. If your LoadBalancer is misconfigured, it is possible that the health is inaccurate due to your misconfiguration. You can also look in https://github.com/argoproj/argo-cd/tree/master/resource_customizations to see how the health is evaluated for each object. In general, a pod should not be ready to accept traffic when it is not present in a load balancer. Argo will not show healthy for a Pod that is not ready.

I would suggest you to open an issue in https://github.com/UCF/UCF-Announcements-Django/issues/new